Abstracts

Rationale: A significant number of elements must be considered in the clinical response to stimulation delivered directly to neuronal populations. These variables include stimulation parameter settings, the number and interdependence of anatomical targets, electrode number, electrode location and orientation, geometry or shape of the electrode contacts, distribution of the cathode and anode, biophysical properties of the stimulated medium, and geometry and trajectory of axonal bundles adjacent to the stimulation site.  A critical step towards applying direct cortical brain stimulation therapy in refractory focal-onset epilepsy is to effectively interface with epileptogenic neural circuits using a small set of active contacts. This has special relevance when interfacing with epileptogenic networks, where two depth leads with four contacts must modulate two or more ictogenic foci.  For this reason, a pre-implant computational model based on compartmental cable modeling was used to predict the interaction of direct neuromodulation of a refractory epileptogenic circuit in a candidate for responsive neurostimulation therapy. Methods: Two depth lead positions were selected in a 3D brain model constructed from a T1 MRI dataset, these positions were designated using subtraction ictal SPECT co-registered with MRI (SISCOM) and parametric subtracted post-ictal DTI. Constrained spherical deconvolution was performed in a high angular resolution diffusion weighted image (60 non-collinear directions, b value= 1000 s/mm²), in conjunction with a probabilistic tractography algorithm. Tracts were generated using seeds that were placed within a spherical region (4mm radius) surrounding the active contacts.  A finite element model was then constructed: An anisotropic conductivity map derived from diffusion tensors was used to solve the electrodynamic quasi-static equations to obtain a time dependent electric potential distribution, produced by applying a bipolar stimulation. Finally, the transmembrane potential for each tract, was calculated using a Hodgkin and Huxley myelinated axon model.  Stereotactic implantation of the depth leads followed our depth electrode placement plan. A post-implant CT was taken to verify the final lead positions. Stimulation activated SPECT (SAS) was used as a technique to visualize transient blood flow changes in response to stimulation without triggering a seizure. Results: The compartmental cable model predicted and differentiated depolarized and hyperpolarized tracts. Moreover, the production of action potentials was modeled in response to the applied stimulus.  Electrocorticographic recordings in conjunction with SAS were used to verify the interconnection of ictal onset zones as predicted by the modulated tractography model.   Conclusions: This simulation effectively predicted the interconnection of two or more ictogenic foci through white matter tracts and their response to the application of direct cortical neurostimulation therapy. It demonstrates the ability to generate a timely pre-implant depth electrode planning map in a refractory epileptogenic network. Funding: CONACYT, Samsung-NeuroLogica